[14:20:36] o/ [15:21:16] O/ [15:31:11] * halfak is going more meta-reviews today. [15:31:17] But hopefully I'll be done in a couple of hours./ [15:31:27] Then i get to work on analysis of new page patrol! [15:31:28] wooo [15:41:55] halfak: want some help with NPP? Im experienced in that area [15:42:51] Zppix, I'm going to be working on bending the database backwards and upside down to recreate history. [15:43:45] You could just query it? [16:00:20] Zppix, haha. That's cute ;) [16:00:29] MediaWiki is an old franken-database [16:00:32] Can't just query it. [16:00:37] halfak: someone on ai mailing list has a question regarding ores models fyi [16:00:42] Zppix, responding [16:01:01] MW doesn't record a bunch of stuff it needs to and has weird anomalies all over the place. [16:01:10] E.g. page moves. [16:01:14] Sounds great, right? [16:01:18] They are in the logging table. [16:01:28] Well the page_id of the page that was actually moved is not in the logging table [16:01:40] So the logging table is useless for figuring out what titles a page used to have. [16:01:52] wow what a night :) [16:06:29] halfak: well if you need me to start reviewing new pages let me know and i can start on enwiki [16:12:39] halfak: is AI team its own or does it have a parent team? [16:14:23] Currently, there isn't an AI team. Starting July 1st (the start of the next fiscal year), I'll be the head of the Scoring Platform team and I'll report directly to the WMF's CTO. [16:15:06] There's more AI going on than just Scoring Platform (where we're focusing on machine learning and prediction models for backlog management) [16:15:24] Zppix, might be interesting to have you help me spot check analysis re. NPP [16:16:08] halfak: okay just let me know [16:16:14] Will do. [16:16:23] halfak: thats my area of expertise when it comes to onwiki xD [16:26:46] halfak: want me to setup weekly ci builds for ores and revscoring to ensure code works? [16:42:01] Dear github, make it possible to have an org wide irc notification setting so i dont have to do it for every single repo manually. with "love" Zppix [16:42:37] wiki-ai/wb-vandalism#167 (CoC - dafe63c : Devin/Zppix): The build passed. https://travis-ci.org/wiki-ai/wb-vandalism/builds/241265540 [16:47:42] rerunning the tests with codecov properly configed fyi [16:52:31] Zppix, not sure that's necessary seeing as we run CI for every commit that gets pushed. [16:52:38] Oh we should kill wb-vandalism [16:52:41] No longer relevant. [16:53:17] ffs okay ill do it just tell me how [16:53:35] I think maybe we should let it live on as Ladsgroup/wb-vandalims [16:53:39] Amir1, what do you think? [16:53:50] Essentially I'm proposing removing it from wiki-ai [16:54:23] ill go ahead and disable travis from that repo (ill keep the config but just turn it off) [16:54:29] kk [16:58:51] im still figuring out why wikibugs isnt here yet, ill keep you updated halfak [17:07:57] ^wait till merging that [17:08:05] i want to talk to Amir1 about that first [17:10:55] Amir1: are you around? [17:33:04] It's okay not to merge! I'm discussing with Aaron right now. [17:36:45] melodykramer: no worries this CoC thing is quite new to me and im unsure about it [17:37:42] I think it was kinda just thrown at us and not explained clearly [17:48:17] Zppix, CoC has been around for a while and discussed a bunch [17:48:51] Don't buy into the trolls who say that this came out of nowhere. They've been involved in the discussion for years and have a minority opinion that we should have any CoC at all. [17:48:52] halfak: i mean bringing into the repos like this [17:48:59] Oh yeah +1 to that [17:49:07] Amir1, is on the committee for the COC I think. [17:49:10] halfak: i know coc was around before [17:49:20] he was involved i know that [17:49:20] Gotcha makes sense then :) [17:49:57] halfak: im going to retry wikibugs by touching a task in scoring team so if you get a notification on phab thats why [17:51:06] yep no luck [18:03:43] Zppix, how come you're pinging me on PRs where I've already finished a review? [18:03:47] E.g. https://github.com/wiki-ai/ores/pull/199 [18:03:54] Waiting on tgr there [18:11:54] i thought it assigned as a reviewer :P (i dont use github besides for wiki ai) :P [18:18:09] Oh I gotcha. Anyway, it's fine. [18:18:23] Just got the ping and thought "Ooh! Was there movement on that PR?" [18:22:15] halfak: doesnt hurt to get a reminder... (as i get notifications for 120 reminders i set on my phone xD) [18:26:36] melodykramer: thanks for understanding its not really your fault, im going to talk with Amir1 and maybe the coc maintainers and see if they can clear up how they want this incorpated into repos cause its not really clear yet [18:26:58] no worries! Always enjoy using GH (and the terminal) so it was nice to head into there for a bit! [18:27:10] Otherwise my skills atrophy. [18:27:49] Also, wanted to bring this event to your attention: http://towcenter.org/artificial-intelligence-practice-and-implications-for-journalism/ - looks like there's a livestream and seems like there might be some interesting talks. [18:27:55] atleast github doesnt fight with you when using terminal (github forces me to use webui ortherwise i get very angry :P) [18:28:14] hmm that is interesting [18:29:02] is that an annual event or something melodykramer ? [18:31:50] I don't think it's annual Zppix - but I know a lot of people who are participating from my previous life in news. [18:32:37] There's also this: https://cyber.harvard.edu/research/ai which just launched and may be f interest. (Announcement from January: https://www.knightfoundation.org/press/releases/knight-foundation-omidyar-network-and-linkedin-founder-reid-hoffman-create-27-million-fund-to-research-artificial-intelligence-for-the-public-interest) [18:33:11] It looks like their symposium is in July: https://artificialintelligencenow.com/schedule/conference [19:14:08] i think i just left the longest comment in phab history :P [21:07:09] hows the npp going halfak ? [21:07:17] Just actually started work [21:07:20] * halfak feels bad [21:07:44] I'm working on generating a dataset that has all page creations since 2008 (when we started keeping track of deleted pages) [21:07:57] just you ? [21:08:23] Then I'll be generating moves historically so that I can find out the status of pages when they were first created. [21:08:54] Then I'll be generating a dataset of creating user statuses -- how long since registration and how many edits saved. [21:11:50] surely to god they arent making use labsdb [21:12:13] thoses things tend to be outdated due to replag issue they are fixing via replacement [21:14:40] i'm not super familiar with it, but finding all the page creations since 2008 sounds like it has overlap with Joal's mediawiki history stuff? Although i know that due to db replica problems it doesn't have all wiki's [21:17:26] halfak: if theres anything i can do to help sift thru the edits or anything ping me [21:18:11] all of those things at least sound like they might be answerable with wmf.mediawiki_page_history and wmf.mediawiki_user_history tables in hive [21:18:30] ebernhardson, certainly does [21:19:07] ebernhardson, not easily still [21:19:13] :( [21:19:16] right :( [21:44:40] halfak: o/ [22:19:05] o/ [22:19:38] If I understand correctly from reading wikiclass repo, I should move the methods that I have created to https://github.com/wiki-ai/wikiclass/blob/master/wikiclass/feature_lists/wikidatawiki.py? [22:20:18] Oh! Looks like I can get one of the hard things from mediawiki_page_history :))) [22:20:20] halfak: I noticed it is different than what I've seen in revscoring. [22:20:26] glorian_wd, yes [22:20:51] "Oh! Looks like I can get one of the hard things from mediawiki_page_history :)))". Did you say this for me? [22:29:31] halfak: could I create a new python file in wikiclass which contains the methods that I created, and call them one by one on wikidatawiki.py?